Compact Ring Main Unit ECT and EVT Multi-Source Data Intelligent Fault Judgment Criteria Method

Abstract

This invention relates to the field of power equipment insulation condition monitoring technology, and discloses a multi-source intelligent fault judgment method for compact ring main units (ECT) and electronic current transformers (EVTs). The method includes: simultaneously acquiring transient voltage and current data of the electronic current transformer in response to a voltage step event triggered by the closing action of the grounding switch; selecting short-time window data at the moment of the step, calculating the ratio of accumulated charge to voltage drop amplitude based on the principle of charge conservation, and calibrating the parasitic capacitance parameters of the equipment; constructing a capacitive displacement component using the parasitic capacitance parameters and voltage change rate, and decoupling it from the total current to reconstruct the dielectric response current reflecting the polarization process of the insulation material; performing multi-timescale decay trend analysis on the dielectric response current to identify the mechanism switching moment, and generating diagnostic results for insulation dampness or interface degradation accordingly. This invention can effectively suppress structural capacitance interference and achieve low-cost, high-precision insulation condition assessment.

Description

Technical Field

[0001] This invention relates to the field of insulation condition monitoring technology for power equipment, and more specifically, to a multi-source data intelligent fault criterion method for compact ring main units (ECT and EVT). Background Technology

[0002] Compact ring main units and gas-insulated switchgear are widely used in medium-voltage power distribution networks due to their small size and strong environmental adaptability. These devices have extremely limited internal space and typically employ a composite insulation structure consisting of gas (such as SF6 or environmentally friendly gases) and solid components (such as epoxy resin insulators). During long-term operation, temperature and humidity cycles and micro-moisture diffusion can cause the solid insulators to absorb moisture or create conductive channels at the interfaces, leading to insulation degradation.

[0003] To meet the needs of intelligent operation and maintenance, these devices are typically equipped with electronic current transformers (ECTs) and electronic voltage transformers (EVTs). However, due to the constraints of their compact physical structure, the measuring elements often need to be integrated with or placed close to detachable cable connectors, and the devices are generally equipped with a permanently online voltage presence indication system (VPIS). This configuration results in a high degree of spatial overlap between the measuring link and the high-field-strength insulation area, and introduces an inherent capacitive coupling network.

[0004] During cable warehouse maintenance or testing, operational standards mandate the mechanical interlocking and closure of the grounding switch. This operation generates a voltage step excitation from energized to ground. The dielectric response current under this excitation contains insulation status information. However, in actual engineering, due to the presence of VPIS and connector parasitic capacitance, an extremely strong and rapid structural displacement current is generated at the moment of grounding. This displacement current, in terms of amplitude and frequency band, often masks the weak polarization / depolarization current within the insulation material caused by moisture or aging.

[0005] Existing insulation monitoring methods mostly rely on single steady-state leakage current or partial discharge monitoring, making it difficult to effectively utilize transient data generated by grounding operations, and unable to extract the true dielectric response from strong structural displacement current interference. This makes it difficult for maintenance personnel to distinguish whether the equipment is in a normal structural charging and discharging state or has experienced a degradation in interface or bulk insulation performance due to moisture. Summary of the Invention

[0006] This invention provides a multi-source data intelligent fault judgment method for compact ring network enclosures using ECT and EVT, which solves the technical problems mentioned in the background art.

[0007] This invention provides a multi-source data intelligent fault judgment method for compact ring network enclosures using ECT and EVT, including:

[0008] When the compact ring network box grounding switch performs a closing operation and generates a voltage step excitation, the transient voltage data output by the electronic voltage transformer and the transient current data output by the electronic current transformer are collected simultaneously.

[0009] By selecting the high-frequency oscillation period after the step jump, and comparing the current integral calculation with the voltage drop amplitude, the parasitic capacitance parameters of the equipment are calibrated to quantify the capacitive coupling effect introduced by the voltage indicator and cable connector.

[0010] The capacitive displacement component is calculated based on the parasitic capacitance parameters and voltage change rate of the device, and the capacitive displacement component is decoupled from the transient current data to reconstruct the dielectric response current that reflects the internal polarization process of the insulating material.

[0011] Multi-timescale decay trend analysis was performed on the dielectric response current to identify the mechanism switching moment of the transition from the interface polarization-dominated stage to the bulk polarization-dominated stage.

[0012] Based on the degree of time drift at the switching moment of the aforementioned mechanism, a state diagnosis result for insulation dampness or interface degradation is generated.

[0013] The beneficial effects of this invention are as follows: Utilizing the grounding switch closing operation occurring during the operation and maintenance of a compact ring main unit as an excitation, and by constructing a charge conservation model to adaptively calibrate the parasitic parameters of the equipment, the invention achieves the removal of structural displacement current interference introduced by voltage indicators and cable connectors, thereby accurately reconstructing the weak dielectric polarization response under strong noise background. Furthermore, this invention eliminates the need for additional sensors, quantifying the dynamic process of the insulating medium transitioning from interface-dominated to bulk-dominated using only existing measurement links. This effectively improves the sensitivity and diagnostic reliability of identifying the moisture absorption and interface degradation states of the internal insulation of the equipment, solving the problem of extracting weak dielectric signals in compact spaces. Attached Figure Description

[0014] Figure 1 This is a flowchart of the intelligent fault judgment method for multi-source data of the compact ring network box ECT and EVT according to the present invention;

[0015] Figure 2 This is a schematic diagram illustrating a specific implementation of the present invention. Detailed Implementation

[0016] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0017] like Figure 1 As shown, the intelligent fault diagnosis method for compact ring main units using ECT and EVT multi-source data includes:

[0018] When the compact ring network box grounding switch performs a closing operation and generates a voltage step excitation, the transient voltage data output by the electronic voltage transformer and the transient current data output by the electronic current transformer are collected simultaneously.

[0019] By selecting the high-frequency oscillation period after the step jump, and comparing the current integral calculation with the voltage drop amplitude, the parasitic capacitance parameters of the equipment are calibrated to quantify the capacitive coupling effect introduced by the voltage indicator and cable connector.

[0020] The capacitive displacement component is calculated based on the parasitic capacitance parameters and voltage change rate of the device, and the capacitive displacement component is decoupled from the transient current data to reconstruct the dielectric response current that reflects the internal polarization process of the insulating material.

[0021] Multi-timescale decay trend analysis was performed on the dielectric response current to identify the mechanism switching moment of the transition from the interface polarization-dominated stage to the bulk polarization-dominated stage.

[0022] Based on the degree of time drift at the switching moment of the aforementioned mechanism, a state diagnosis result for insulation dampness or interface degradation is generated.

[0023] In a preferred embodiment, when the compact ring main unit grounding switch performs a closing operation and generates a voltage step excitation, the transient voltage data output by the electronic voltage transformer and the transient current data output by the electronic current transformer are simultaneously acquired, including:

[0024] The Singer reconstruction filtering algorithm is used to resample the original discrete voltage sequence output by the electronic voltage transformer and the original discrete current sequence output by the electronic current transformer to an internally unified sampling rate. Time-aligned transient voltage data were obtained. With transient current data The calculation formula is as follows:

[0025]

[0026] in, This refers to the resampled transient voltage or transient current data. This is the original collected data. For normalized time axis;

[0027] For the transient voltage data The numerical voltage derivative sequence is obtained by performing a second-order central difference operation. And search its maximum index to determine the ground step trigger time. The calculation formula is as follows:

[0028]

[0029]

[0030] in, This is the index of the sampling point where the absolute value of the numerical voltage derivative sequence is the largest.

[0031] When a compact ring main unit grounding switch performs a closing operation that generates a voltage step excitation, the specific implementation of simultaneously acquiring transient voltage data output by an electronic voltage transformer and transient current data output by an electronic current transformer requires addressing the sampling rate differences and timestamp misalignments inherent in the original sampling systems of different transformers. Considering the potential differences in hardware design between electronic voltage and current transformers, their original discrete voltage and current sequences often have inconsistent sampling frequencies. Direct use of these sequences can lead to distortion in the subsequent time correspondence between current and voltage. Therefore, a unified resampling mechanism is needed. The Singer reconstruction filtering algorithm possesses the characteristic of maintaining signal spectral integrity during arbitrary sampling rate conversions. This algorithm resamples both types of original data to an internally unified sampling rate. It can achieve precise timestamp alignment, and the expression for the resampling process is: ,in This represents the resampled transient voltage or transient current data. This is the original collected data. To normalize the time axis, this expression uses the interpolation properties of the Singer function to ensure that the resampled data can accurately reproduce the amplitude and phase characteristics of the original signal.

[0032] After resampling, the grounding step trigger time needs to be locked. A characteristic of the voltage step process is that the voltage change rate reaches an extreme value within a short period; therefore, this feature can be extracted by performing differential operations on the transient voltage data. A second-order central difference algorithm is chosen to calculate the numerical voltage derivative sequence, whose expression is: This algorithm calculates the derivative using the voltage values ​​of three adjacent sampling points. Compared to the first-order difference algorithm, it effectively reduces noise interference while maintaining a balance between the accuracy of the derivative calculation and the time delay, resulting in a more accurate and efficient derivative. It can accurately reflect the rate of voltage change over time. Since the voltage step caused by the closing of the grounding switch is a rapid drop in voltage from steady state to zero, the absolute value of the numerical voltage derivative sequence reaches its maximum value at this time. Therefore, the moment corresponding to this maximum value is locked as the grounding step trigger moment. Its corresponding index satisfy , .

[0033] Finally, based on the determined ground step trigger time To ensure the comprehensiveness of subsequent analysis, a data window containing the complete process needs to be extracted. The voltage and current response after the grounding switch is closed comprises three key stages: the steady-state period before the step jump, the transient period after the step jump, and the dielectric response period after the step jump. The voltage data during the steady-state period before the step jump is used to calculate the stable amplitude before the step jump. The transient period includes the superimposed signal of structural displacement current and initial polarization current. The dielectric response period after the step jump mainly reflects the polarization decay process of the insulating material. Therefore, the data window needs to be extracted based on... Centered on the step, the data extends forward to a sufficiently long steady-state interval before the step and backward to cover the effective duration of the step transient process and the dielectric response, ensuring that the captured dataset can completely contain all electrical response information under ground step excitation.

[0034] In a preferred embodiment, the high-frequency oscillation period following the step change is selected, and the parasitic capacitance parameters of the device are calibrated by comparing the current integration calculation with the voltage drop amplitude, in order to quantify the capacitive coupling effect introduced by the voltage indicator and cable connector, including:

[0035] Based on the ground step trigger time Calculate transient voltage data voltage step amplitude The calculation formula is as follows:

[0036]

[0037]

[0038] in, The steady-state mean before the step jump. This is the steady-state mean after the step jump. For the steady-state calculation window before the step jump, To avoid the steady-state calculation window after the ringing step, , These represent the number of data points in the corresponding window;

[0039] Transient current data Intercepting the charge integration time window The cumulative charge is calculated by discrete integration. The calculation formula is as follows:

[0040]

[0041] in, The short time window length is dominated by displacement current. To ensure a uniform sampling rate internally, This is the index corresponding to the grounding step trigger moment;

[0042] Calculate the accumulated charge. With voltage step amplitude The ratio of the two values ​​yields the parasitic capacitance parameters of the equipment. :

[0043]

[0044] The parasitic capacitance parameters of the device Used as displacement current for structural fingerprint stripping in subsequent steps.

[0045] In selecting the high-frequency oscillation period after the step jump, and comparing the current integration calculation with the voltage drop amplitude to calibrate the parasitic capacitance parameters of the equipment, the primary task is to obtain the voltage step amplitude. The voltage step caused by the closing of the grounding switch is accompanied by mechanical ringing. The voltage data during the ringing period is affected by transient interference and cannot reflect the true steady-state voltage level. Therefore, it is necessary to select a steady-state calculation window before the step. and the steady-state calculation window after avoiding ringing step Steady-state calculation window before step jump Located at the ground step trigger time Previously, the voltage during this period was unaffected by the step change and remained in a stable state, based on the transient voltage data within the window. Taking the average value yields the steady-state average value before the step jump. Its calculation expression is: ,in for The number of data points within the range ensures that the mean accurately reflects the voltage reference before the step jump. The steady-state calculation window after the step jump... The starting point needs to span the mechanical ringing decay period. Data should be extracted after the voltage stabilizes, and the steady-state mean after the step jump should be obtained through averaging. The expression is , for The number of data points within, through and The voltage step amplitude can be obtained from the difference. This eliminates the interference of ringing on voltage amplitude calculation.

[0046] Get Then, the accumulated charge corresponding to the parasitic capacitance needs to be calculated. The parasitic capacitance of the compact ring main unit is mainly introduced by the voltage indicator and cable connectors. Its charge release process is concentrated in a short time window after the grounding step, during which displacement current dominates and the influence of polarization current is negligible. Therefore, it is necessary to select a charge integration time window that is close to the grounding step trigger moment and is in the charge accumulation plateau period. The charge accumulation plateau period signifies that the discharge of charge from the parasitic capacitance is essentially complete, at which point transient current data... By performing cumulative integration, the total charge stored in the parasitic capacitance can be completely captured. The expression for the integration operation is: ,in This is the index corresponding to the ground step trigger moment. To ensure a uniform sampling rate internally, The time window length is converted into the number of sampling points to ensure that the integration range accurately covers the entire process of charge release.

[0047] The calculation of parasitic capacitance is based on the fundamental relationship of capacitance, namely, the capacitance value is equal to the ratio of charge to voltage. Therefore, the accumulated charge is... With voltage step amplitude Dividing by this yields the parasitic capacitance parameters of the device. This parameter comprehensively characterizes the equivalent capacitance network formed by the voltage indicator and cable connector inside the compact ring main unit in a high field strength region. Its calibration process fully utilizes the transient electrical characteristics after the grounding step and eliminates interference from polarization current and mechanical ringing.

[0048] In a preferred embodiment, calculating the capacitive displacement component based on the device's parasitic capacitance parameters and voltage change rate includes:

[0049] Using the parasitic capacitance parameters of the device Numerical voltage derivative sequence with transient voltage data Construct capacitive displacement components The calculation formula is as follows:

[0050]

[0051] in, For capacitive displacement components, This is a numerical voltage derivative sequence. The parameters are the parasitic capacitance of the equipment.

[0052] The capacitive displacement component Used to subtract structural disturbances from the total current in subsequent steps.

[0053] The process of calculating the capacitive displacement component based on the parasitic capacitance parameters and voltage change rate of the equipment involves constructing the current component corresponding to the inherent capacitor network based on the fundamental conduction relationship of capacitor current. Equipment parasitic capacitance parameters This comprehensively reflects the characteristics of the equivalent capacitance network formed by the voltage indicator and cable connector in the high field strength region, while the numerical voltage derivative sequence obtained from the transient voltage data through second-order central difference operation... This accurately captures the rate of voltage change over time during a grounding step, and the combination of these two factors reconstructs the capacitive displacement component. The displacement current generated by the capacitor network during voltage changes is related to both the capacitance value and the rate of voltage change. This relationship is expressed as a point-by-point product in the discrete-time domain. Therefore, the parasitic capacitance parameters of the equipment... With numerical voltage derivative sequence By multiplying each point individually, the capacitive displacement components can be obtained. The calculation process follows the conduction law of capacitive current. The capacitive displacement component at each sampling moment corresponds to the non-polarized conduction current generated by the voltage change in the inherent capacitor network at that moment. Its essence is the charging and discharging current of the structural capacitor, which is unrelated to the polarization process of the insulating material.

[0054] In a preferred embodiment, the capacitive displacement component is decoupled from the transient current data to reconstruct the dielectric response current reflecting the internal polarization process of the insulating material, including:

[0055] The capacitive displacement component From transient current data The polarization wake current was obtained by stripping the middle layer. The calculation formula is as follows:

[0056]

[0057] Using the voltage step amplitude value polarization wake current Normalization is performed to reconstruct the dielectric response current. The calculation formula is as follows:

[0058]

[0059] in, The dielectric response current (i.e., the normalized admittance kernel function) is used to characterize the dynamic response characteristics of the insulating dielectric under unit voltage excitation.

[0060] The capacitive displacement component is decoupled from the transient current data and the dielectric response current is reconstructed to separate structural interference from insulation polarization signals. Then, standardization is used to eliminate differences in external conditions. (Capacitive displacement component) It relates only to the charging and discharging of the device's inherent capacitor network, does not contain any polarization information of insulating materials, and synchronously collects transient current data. It is a superposition signal of capacitive displacement component and polarization current, therefore, point-by-point differential operation is required to separate the two types of signals. The expression for differential operation is: This operation removes the influence of structurally non-polarized conduction currents from the total current, leaving only the polarization wake current. It only carries the current changes generated inside the insulating material due to the polarization process.

[0061] obtain polarization wake current Subsequently, normalization processing is required to eliminate the impact of voltage level differences. Different compact ring main units may be configured with different rated voltages, and the voltage step amplitude of the same equipment under different operation and maintenance scenarios may vary. Fluctuations may also exist; these differences in external conditions can lead to variations in the amplitude range of the polarization wake current, resulting in deviations from direct comparisons. Therefore, the polarization wake current... Divide by voltage step amplitude The normalized admittance kernel function is obtained. This function is essentially the dynamic response of the insulating medium under unit voltage excitation. Its amplitude and trend are determined solely by the polarization characteristics of the insulating material itself, and are independent of external voltage conditions. Through this standardization process, dielectric response data under different voltage levels and different step amplitudes have a comparable basis. Finally, the normalized admittance kernel function is determined as the dielectric response current. .

[0062] In a preferred embodiment, multi-timescale decay trend analysis is performed on the dielectric response current to identify the mechanism switching moments from the interface polarization-dominated stage to the bulk polarization-dominated stage, including:

[0063] Define relative time In the logarithmic field Below, the current responds to the medium Perform piecewise linear fitting;

[0064] In the interface polarization analysis window Inside, the fitting line for the interface layer is obtained:

[0065]

[0066] In bulk polarization analysis window Within, the fitted line for the volumetric layer is obtained:

[0067]

[0068] in, The intercept is... The polarization memory index;

[0069] Calculate the intersection of the two lines to obtain the timing of the mechanism switching. :

[0070]

[0071] The mechanism switching time The dynamic equilibrium point used to characterize the layered polarization process inside the insulating medium.

[0072] To analyze the decay trend of the dielectric response current across multiple time scales to identify the mechanism switching moment, a unified time reference and a suitable analysis coordinate system must first be established. This is because the dielectric response current... Using absolute time as a reference, the grounding step trigger time It is the starting node for all responses, therefore relative time is defined. This approach unifies the response data from different events onto a time axis with the step time as the origin, eliminating analytical biases caused by absolute time differences. The decay process of the dielectric response current exhibits nonlinear characteristics, making it difficult to distinguish the dominant stages of different polarization mechanisms through direct analysis. Logarithmic transformation can convert nonlinear decay into a linear relationship; therefore, a logarithmic coordinate system is constructed. , This makes the decay trends of interface polarization and volume polarization appear as straight lines in the coordinate system, providing a basis for piecewise fitting.

[0073] Based on the polarization characteristics of the insulating medium, the polarization process after a grounding step exhibits distinct stages: the early stage is dominated by interface polarization, characterized by a fast response and steep attenuation; the later stage is dominated by bulk layer polarization, with a slow response and gradual attenuation. The transition between these two stages marks the mechanism switching point. Therefore, a corresponding analysis window needs to be selected in the logarithmic coordinate system, specifically the interface polarization analysis window. Covering the early period after the step change ensures that the data within the window only reflects the interface polarization decay characteristics; volume polarization analysis window To cover the late-stage period, the data within each window reflects only the polarization decay characteristics of the bulk layer. Linear least-squares fitting is performed on the data points within both analysis windows to obtain the interface layer fitting line. Fitting straight line with volumetric layer ,in , The logarithmic intercept corresponds to the initial amplitude level of the polarization response. , The polarization memory index quantifies the decay rate of two types of polarization.

[0074] The mechanism switching moment corresponds to the point at which the contributions of the two polarization mechanisms are equal, which is represented in a logarithmic coordinate system as the intersection of two fitted lines. Let the equations of the two lines be equal. After rearranging, we can obtain the x-coordinate of the intersection point. The horizontal axis corresponds to the logarithmic form of relative time. Therefore, the mechanism switching time can be obtained through exponential transformation. This value directly corresponds to the timescale of the transition of the insulating medium from interface-dominated polarization to bulk-dominated polarization.

[0075] In a preferred embodiment, a condition diagnosis result for insulation dampness or interface degradation is generated based on the degree of time drift at the switching moment of the mechanism, including:

[0076] Using the interface layer to fit the slope of a straight line The slope of the fitted line of the volumetric layer and the timing of mechanism switching Constructing hierarchical complexity strength The calculation formula is as follows:

[0077]

[0078] in, The maximum observation time constant;

[0079] Using the logistic mapping function to reduce the complexity of hierarchical structures Convert to fault criterion value The calculation formula is as follows:

[0080]

[0081] in, For the slope parameter, The center offset parameter;

[0082] Binary state diagnosis results are generated using a step function. :

[0083]

[0084] in, This is the alarm threshold. Let Heviside's step function be used; when The condition is determined to be due to insulation dampness or interface deterioration.

[0085] The process of generating diagnostic results for insulation moisture or interface degradation based on the degree of time drift at the mechanism switching moment first requires constructing a layered complexity strength that can comprehensively reflect changes in insulation state. Due to the slope of the fitted line of the interface layer The slope of the straight line fitted to the volumetric layer The magnitude of this difference is directly related to the degree of separation between the two types of polarization attenuation characteristics. Insulation dampness or interface degradation will significantly increase this difference, while the square operation of the slope difference can effectively amplify this change and eliminate the influence of positive and negative signs; mechanism switching time The time drift directly reflects the pace of the shift in the polarization-dominant mechanism. A smaller value indicates that the interface polarization-dominated phase ends earlier and the insulation state anomaly is more pronounced, which, combined with the maximum observation time... and Taking the ratio and its logarithm will give us... The drift is transformed into a continuous and quantifiable value, thus a hierarchical complexity strength is constructed by multiplying the two. This enables a comprehensive characterization of changes in insulation state.

[0086] Obtain the strength of hierarchical complexity Then, it needs to be mapped to a normalized fault criterion value. This is to facilitate the setting of uniform alarm thresholds for judgment. The numerical range of these values ​​can fluctuate widely depending on the insulation state of the equipment, making it difficult to establish a unified standard using them directly. However, the logistic mapping function can map any real number to... The interval characteristic allows different degrees of insulation anomalies to be mapped to values ​​within a fixed range, and the mapping process is continuous and smooth, avoiding misjudgments caused by abrupt numerical changes; the slope parameter in the function With center offset parameter The calibration was performed offline, based on a large amount of sample data with different insulation conditions, to ensure... It can reasonably control the steepness of the mapping curve. It can accurately correspond to the hierarchical complexity strength benchmark of health equipment, and finally obtain the fault criterion value. .

[0087] Finally, by comparing with the preset alarm threshold The comparison generates a binary diagnostic result. The calibration process, conducted offline, comprehensively considered the equipment's operational safety requirements and actual maintenance data, balancing false alarm and detection rates to ensure accurate delineation of normal and abnormal boundaries. The Heaviside step function was employed. Make a judgment when the fault criterion value Exceed When, the function outputs This indicates that the insulation has become damp or the interface has deteriorated; when Not exceeding When, the function outputs This indicates that the insulation is in normal condition, making it easier for maintenance personnel to quickly identify the equipment status.

[0088] The raw discrete voltage sequence refers to the unprocessed discrete voltage data directly output by the electronic voltage transformer, serving as the fundamental input for subsequent operations such as resampling and derivative calculation. The sampling rate of this sequence is determined by the transformer's hardware design and may vary between different devices. It can be directly acquired in real time through the device's legitimate and compliant hardware interface.

[0089] The raw discrete current sequence refers to the unprocessed discrete current data directly output by the electronic current transformer, which is acquired synchronously with the raw discrete voltage sequence during the voltage step event of the grounding switch closing. Its sampling rate is also determined by the transformer hardware and may differ from the raw discrete voltage sequence; it can be obtained in real time through the device hardware interface.

[0090] The Singer reconstruction filter algorithm is an algorithm used to unify data with different sampling rates, through the Singer function. Interpolation operations are performed to maintain the spectral integrity of the original signal and avoid signal distortion during sampling rate conversion. The Singer reconstruction filter algorithm is used to resample the original discrete voltage sequence and the original discrete current sequence to the same internal uniform sampling rate.

[0091] Internal uniform sampling rate This refers to the reference parameter used to standardize the sampling frequency of all data. It is obtained through offline calibration: spectral analysis is performed on short-window data near the step of all ground step events to calculate the 99% energy frequency. Get all events The maximum value, multiplied by the oversampling coefficient (Verified offline) (This balances noise and accuracy), and is then rounded to the commonly used engineering standard value. The preferred value is 100kHz.

[0092] Raw data collection The input data refers to the Singh reconstruction filter algorithm. This is a general identifier that can refer to either the original discrete voltage sequence or the original discrete current sequence. This is the sampling index for the raw data. Raw collected data. It is generated naturally with the collected raw data and is the data source for resampling operations.

[0093] Normalized time axis The time axis used to standardize the time base for resampled data is calculated as follows: ,in This is the sampling index after resampling. This ensures a unified internal sampling rate. This timeline maps data from different original sampling rates to the same time scale, guaranteeing accurate timestamp alignment between resampled transient voltage and transient current data.

[0094] Resampled data This refers to the standardized data obtained after processing by the Singer reconstruction filter algorithm. This is a general identifier that can refer to transient voltage or transient current data. This is the sampling index after resampling.

[0095] Transient voltage data This refers to time-aligned voltage data obtained by resampling the original discrete voltage sequence using the Singer reconstruction filter algorithm. This is the sampling index after resampling. This data has eliminated the original sampling rate differences and is used to calculate the numerical voltage derivative sequence. and voltage step amplitude .

[0096] Transient current data This refers to time-aligned current data obtained by resampling the original discrete current sequence using the Singer reconstruction filter algorithm. This is the sampling index after resampling. Transient current data. With transient voltage data Timestamp synchronization.

[0097] Numerical voltage derivative sequence It refers to a discrete sequence that reflects the rate of change of voltage over time. This is the sampling index after resampling. The second-order central difference algorithm can reduce noise interference while ensuring computational accuracy, accurately capturing rapid voltage changes. Numerical voltage derivative sequence. Used to lock the trigger time of a grounding step, for example, when the grounding switch closes, causing a rapid voltage drop. Values ​​with significantly increased absolute values ​​will appear.

[0098] The index corresponding to the step trigger time Exponential voltage derivative sequence The index of the sampling point where the absolute value is maximum. This index is dynamically determined as the transient voltage data changes, mapping the extreme points of the derivative sequence to actual time (the grounding step trigger moment). (a key intermediary)

[0099] Grounding step trigger time The time point at which the voltage undergoes a step change due to the closing of the grounding switch.

[0100] The complete data window refers to the time based on the ground step trigger. The extracted data set contains three key time periods, namely the pre-step steady-state period ( The period of stable voltage before the step transient period ( The period of rapid voltage change before and after the step, and the period of dielectric response after the step ( (The period of subsequent polarization current decay). This window has no fixed duration and needs to be truncated based on the actual response characteristics of the equipment to ensure coverage of the complete electrical response process after the grounding step. For example, it can be truncated... to The time period includes both the steady state before the step and the transient and polarization response after the step.

[0101] Pre-step steady-state calculation window Refers to the time when the ground step is triggered. The previous voltage data window only included periods of stable voltage, used to calculate the steady-state mean before the step jump. The window's position should avoid any transient interference; a preferred value is [value to be filled in]. (i.e., relative to) of to The interval is determined through offline calibration (avoiding possible minor fluctuations before a step jump).

[0102] Step-start steady-state calculation window Refers to the time when the ground step is triggered. The subsequent voltage data window should begin only during periods when the voltage is stabilizing, avoiding the mechanical ringing decay phase, and be used to calculate the steady-state mean after the step jump. The preferred value for this window is... (i.e., relative to) of to The interval was determined through offline calibration (the 95th percentile of the ringing end time).

[0103] steady-state mean before step Pre-step steady-state calculation window Internal transient voltage data The arithmetic mean. Steady-state mean before step. Used to characterize the stable voltage level of a device before a grounding step occurs.

[0104] Steady-state mean after step The steady-state calculation window after a step jump Internal transient voltage data The arithmetic mean. Steady-state mean after a step jump. Used to characterize the voltage level when it stabilizes after a grounding step occurs.

[0105] Voltage step amplitude This refers to the core parameter reflecting the voltage change amplitude during a ground fault step event. Voltage step amplitude. It changes dynamically with the operating voltage and step effect of the equipment.

[0106] Charge integration time window The moment immediately following the grounding step trigger. Furthermore, a short-time current data window during the charge accumulation plateau period is used to fully capture all the charge released by the device's parasitic capacitance. The preferred value for this window is 2ms, determined through offline calibration: calculate the arrival time of the charge accumulation plateau for all events, take its 95th percentile, and round it to an engineering-friendly value.

[0107] Accumulated charge Charge integration time window Internal transient current data The result of the cumulative integral. Accumulated charge. Used to characterize the total charge stored in the parasitic capacitance of a device.

[0108] Equipment parasitic capacitance parameters These parameters characterize the equivalent capacitance network formed by voltage indicators, cable connectors, and parasitic capacitances within a compact ring main unit. (Equipment parasitic capacitance parameters) Dynamic calibration based on equipment structure and operating status enables the quantification of capacitive coupling effects.

[0109] capacitive displacement components This refers to the sequence of non-polarized conduction currents generated during a grounding step by the inherent capacitance network within the equipment (voltage indicators, cable connectors, and parasitic capacitances). This is the sampling index after resampling. Capacitive displacement components. It is unrelated to the polarization process of the insulating medium and only reflects the charging and discharging characteristics of the structural capacitor. It is an interference item that needs to be removed from the total current.

[0110] Polarization wake current This refers to a current sequence that contains only polarization information of the insulating medium. This is the sampling index after resampling. After removing structural interference through point-by-point differential operation, this current sequence only reflects the polarization / depolarization process inside the insulating material.

[0111] Normalized admittance kernel function (dielectric response current) The dynamic response parameters of the medium after eliminating the influence of voltage level. This is the sampling index after resampling, in Siemens (S). This normalization process makes the dielectric response data under different voltage levels and different step amplitudes comparable.

[0112] relative time Refers to the ground step trigger time A time reference established at the origin. Relative time. This is used to unify the time scale of different grounding step events, eliminate the influence of absolute time differences on the analysis results, and enable direct comparison of the polarization responses of different events.

[0113] logarithmic coordinate system x-axis The horizontal axis is used to convert nonlinear polarization decay into a linear relationship. Since the decay of dielectric polarization current exhibits a power-law characteristic, it is difficult to distinguish different polarization mechanisms through direct analysis. Logarithmic transformation can convert power-law decay into a linear relationship, facilitating piecewise fitting.

[0114] Logarithmic coordinate system vertical axis The vertical axis is used to transform nonlinear polarization attenuation into a linear relationship. Combined with a logarithmic transformation of the horizontal axis, this transformation allows the attenuation trends of interface polarization and bulk layer polarization to appear as straight lines.

[0115] Interface polarization analysis window This refers to a data analysis window located in the relatively early time frame, covering only the period dominated by interface polarization, used to extract the decay characteristics of interface polarization. The preferred value for this window is... Determined through offline calibration: maximizing the median of logarithmic domain linear regression within the candidate window set. (With window width penalty) to ensure that the data within the window only reflects the interface polarization process.

[0116] Volume polarization analysis window This refers to a data analysis window located in the relatively late time period, covering only the time period dominated by bulk layer polarization, used to extract the decay characteristics of bulk layer polarization. The preferred value for this window is... Determined through offline calibration: maximizing the median of logarithmic domain linear regression within the candidate window set. (with window width penalty) to ensure that the data within the window only reflects the volumetric polarization process.

[0117] Interface layer fitted linear intercept The logarithmic intercept, obtained by fitting data within the interface polarization analysis window using linear least squares, is dynamically generated based on the interface polarization decay characteristics. This is the intercept of the fitted linear line at the interface layer. It is related to the initial amplitude level of the interface polarization response.

[0118] Volumetric layer fitting linear intercept The logarithmic intercept of the data within the volume polarization analysis window, obtained by linear least squares fitting, is dynamically generated based on the volumetric layer polarization attenuation characteristics. The intercept of the fitted linear sequence for the volumetric layer is also shown. It is related to the initial amplitude level of the bulk layer polarization response.

[0119] Interface layer polarization memory index The absolute value of the slope of the fitted line of the interface layer, with a range of values. The polarization memory index of the interface layer was obtained through linear least squares fitting. This is used to quantify the decay rate of interface polarization; a larger value indicates that the interface polarization decays faster.

[0120] Bulk phase layer polarization memory index The absolute value of the slope of the fitted straight line of the phase layer refers to the range of values. The volumetric layer polarization memory index was obtained through linear least squares fitting. This is used to quantify the decay rate of bulk layer polarization; a larger value indicates that the bulk layer polarization decays faster.

[0121] Mechanism switching time The timescale for the transition of the insulating medium from the interfacial polarization-dominated stage to the bulk layer polarization-dominated stage is a core, specific parameter characterizing the insulation state. Mechanism switching time. As the insulating medium becomes damp and deteriorates, it dynamically shifts, and a systematic shift will occur when the insulation is abnormal.

[0122] Maximum observation time constant Refers to the strength of layered complexity. The time parameter, preferably 5s, is determined through offline calibration: the signal-to-noise ratio of the wake falling below the noise floor is calculated for all events. The available termination time is taken as the 90th percentile and rounded to the nearest engineering-friendly value. Maximum observation time constant. Ensure the effective time period of coverage polarization response, for The calculation provides a stable time reference.

[0123] Layered complexity strength It refers to a parameter that comprehensively reflects the degree of difference in the layering of the insulating medium, integrating the differences in polarization characteristics between the interface layer and the bulk layer, as well as the drift information at the moment of mechanism switching. reflect The degree of drift. When insulation is abnormal. It will increase significantly.

[0124] slope parameter The steepness control parameter of the logistic mapping function is determined through offline calibration: with the healthy state (S0) as label 0 and the deteriorated state (S2) as label 1, the following parameters are applied: Logistic regression was performed on the labels to obtain a unique slope. ,Right now The preferred value for this parameter is 8, ensuring... Tiny changes can be converted into fault criterion values. Achieving a significant response while avoiding oversensitivity.

[0125] Center offset parameter The reference center parameter of the logistic mapping function is determined through offline calibration: the intercept obtained from logistic regression fitting. ,calculate ( (This is the fitting slope), corresponding to the hierarchical complexity strength benchmark of the health device. The preferred value for this parameter is 0.35, ensuring the health device... Concentrated in the low range, deteriorating equipment Concentrated in the high range.

[0126] Fault criterion value This refers to the normalized insulation condition risk value, with a range of values ​​including: Fault criterion value Using logistic mapping The risk value is converted into a uniform scale. The closer it is to 1, the higher the risk of insulation moisture or interface deterioration. The closer it is to 0, the healthier the insulation condition.

[0127] alarm threshold This refers to the boundary parameter that distinguishes between normal and abnormal insulation conditions, determined through offline calibration: Based on the validation set (including states S0, S1, and S2), the ROC curve is calculated, a target false alarm rate (e.g., 1%) is set, and the minimum threshold that meets the false alarm rate requirement is selected. Alarm Threshold The preferred value is 0.8, when If the insulation is faulty, it is considered an insulation abnormality; otherwise, it is considered normal.

[0128] Hevyside step function This refers to a fixed logic function used to generate binary diagnostic results, defined as: outputting 1 when the input value is greater than or equal to 0, and outputting 0 when the input value is less than 0. This function has no additional adjustable parameters; it only serves as a carrier for the judgment logic, transmitting the fault criterion value... With alarm threshold The comparison results are converted into explicit binary labels.

[0129] Binarization status diagnosis results This refers to the final fault determination label, with a value of [value missing]. . This indicates that the insulation is damp or the interface is deteriorated. This indicates that the insulation condition is normal; the binarized state diagnosis result is as follows. It makes it easier for maintenance personnel to quickly identify the status of equipment.

[0130] like Figure 2 As shown, Figure 2This is a schematic diagram of the structural principle of a compact ring main unit (RMU) intelligent fault diagnosis method, demonstrating the hardware layout that uses the closing of a grounding switch as an excitation source to detect the insulation condition. The upper half of the diagram represents the switch compartment, depicting the instant the grounding switch is closing, a mechanical action that generates a voltage step excitation within the system. The lower half represents the cable compartment, where electronic voltage transformers (EVTs) and electronic current transformers (ECTs) are arranged around the main body to synchronously acquire the transient voltage and current data generated. High-field-strength areas are outlined with dashed lines, and the equivalent parasitic capacitance network introduced by the voltage indicator and cable connectors is represented by capacitance symbols. The acquired multi-source data is converged via signal lines to the data processing unit on the right, used for subsequent calculation of capacitive displacement components and extraction of the dielectric response current reflecting insulation moisture or aging.

[0131] It is important to note that all input data described in this solution is acquired in real-time through legal and compliant hardware interfaces with the user's full knowledge, explicit consent, and active cooperation. The preset parameters, prior constants, and statistical means are all derived from publicly available scientific literature data, de-identified general research datasets, or calibration data from laboratory environments, and do not contain any unauthorized sensitive third-party information. The system's data processing is limited to local or volatile memory computation transmitted via encrypted channels. There is no illegal collection, theft, or retention of user biometric data or infringement of user privacy without the user's knowledge. All parameter calls and generation comply with the principles of data minimization, legality, legitimacy, and necessity.

[0132] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A multi-source data intelligent fault judgment method for compact ring main units using ECT and EVT, characterized in that, include: When the compact ring network box grounding switch performs a closing operation and generates a voltage step excitation, the transient voltage data output by the electronic voltage transformer and the transient current data output by the electronic current transformer are collected simultaneously. By selecting the high-frequency oscillation period after the step jump, and comparing the current integral calculation with the voltage drop amplitude, the parasitic capacitance parameters of the equipment are calibrated to quantify the capacitive coupling effect introduced by the voltage indicator and cable connector. The capacitive displacement component is calculated based on the parasitic capacitance parameters and voltage change rate of the device, and the capacitive displacement component is decoupled from the transient current data to reconstruct the dielectric response current that reflects the internal polarization process of the insulating material. Multi-timescale decay trend analysis was performed on the dielectric response current to identify the mechanism switching moment of the transition from the interface polarization-dominated stage to the bulk polarization-dominated stage. Based on the degree of time drift at the switching moment of the aforementioned mechanism, a state diagnosis result for insulation dampness or interface degradation is generated.

2. The intelligent fault judgment method for compact ring network box ECT and EVT multi-source data according to claim 1, characterized in that, When the compact ring main unit grounding switch performs a closing operation and generates a voltage step excitation, the transient voltage data output by the electronic voltage transformer and the transient current data output by the electronic current transformer are simultaneously collected, including: The original discrete voltage sequence output by the electronic voltage transformer and the original discrete current sequence output by the electronic current transformer are uniformly resampled to an internal uniform sampling rate using the Singer reconstruction filtering algorithm to obtain time-stamp aligned transient voltage and transient current data. The transient voltage data is subjected to a second-order central difference operation to obtain a numerical voltage derivative sequence, and the moment with the largest absolute value in the numerical voltage derivative sequence is locked as the ground step trigger moment. Based on the grounding step trigger time, a complete data window is extracted, including the steady-state period before the step, the transient period of the step, and the dielectric response period after the step.

3. The intelligent fault judgment method for compact ring network box ECT and EVT multi-source data according to claim 1, characterized in that, By selecting the high-frequency oscillation period following the step jump, and comparing the current integration calculation with the voltage drop amplitude, the parasitic capacitance parameters of the equipment are calibrated to quantify the capacitive coupling effect introduced by the voltage indicator and cable connector, including: In the transient voltage data, a steady-state calculation window before the ground step trigger moment and a steady-state calculation window after the ground step trigger moment are selected respectively. The difference between the mean values ​​of the steady-state calculation window before the ground step trigger moment and the steady-state calculation window after the ground step trigger moment are calculated as the voltage step amplitude. The starting time of the steady-state calculation window after the ground step trigger moment should avoid the mechanical ringing decay period. In the transient current data, a charge integration time window is selected that is immediately adjacent to the grounding step trigger time and is in the charge accumulation plateau period. The transient current data within the charge integration time window is accumulated and integrated to obtain the accumulated charge. Calculate the ratio of the accumulated charge to the voltage step amplitude, and use this ratio as the parasitic capacitance parameter of the equipment to characterize the equivalent capacitance network formed by the voltage indicator and cable connector inside the compact ring main unit in the high field strength region.

4. The intelligent fault judgment method for compact ring network box ECT and EVT multi-source data according to claim 3, characterized in that, The capacitive displacement component is calculated based on the parasitic capacitance parameters and voltage change rate of the device, including: The numerical voltage derivative sequence obtained after differential operation of transient voltage data is acquired is then multiplied point by point with the parasitic capacitance parameter of the device. The product sequence obtained from the calculation is taken as the capacitive displacement component, which represents the non-polarized conduction current generated by the inherent capacitance network inside the equipment during the closing process of the grounding switch.

5. The intelligent fault judgment method for compact ring network box ECT and EVT multi-source data according to claim 1, characterized in that, By decoupling the capacitive displacement component from the transient current data, the dielectric response current reflecting the internal polarization process of the insulating material is reconstructed, including: Perform differential denoising operation to subtract the calculated capacitive displacement component from the synchronously acquired transient current data point by point to obtain the polarization wake current containing only dielectric polarization information. Normalization is performed by dividing the polarization wake current by the voltage step amplitude to obtain a normalized admittance kernel function that eliminates the influence of voltage levels, and this normalized admittance kernel function is used as the dielectric response current.

6. The intelligent fault judgment method for compact ring network box ECT and EVT multi-source data according to claim 1, characterized in that, Multi-timescale decay trend analysis was performed on the dielectric response current to identify the mechanism switching moments from the interface polarization-dominated stage to the bulk polarization-dominated stage, including: A logarithmic coordinate system is constructed with relative time as the horizontal axis and the dielectric response current amplitude as the vertical axis. In the logarithmic coordinate system, the interface polarization analysis window located in the early part of the time axis and the volume polarization analysis window located in the late part of the time axis are selected. Linear least squares fitting was performed on the data points within the interface polarization analysis window and the volume polarization analysis window respectively to obtain the interface layer fitting line characterizing the polarization attenuation characteristics of the interface layer and the volume layer fitting line characterizing the polarization attenuation characteristics of the volume phase layer. The intersection point of the fitted line of the interface layer and the fitted line of the bulk layer is calculated, and the value corresponding to the intersection point on the time axis is taken as the mechanism switching time. The mechanism switching time characterizes the time scale of the insulating medium switching from the interface-dominated polarization process to the bulk-dominated polarization process.

7. The intelligent fault judgment method for compact ring main unit ECT and EVT multi-source data according to claim 6, characterized in that, Based on the degree of time drift at the switching moment of the aforementioned mechanism, a state diagnosis result for insulation moisture or interface degradation is generated, including: The square of the slope difference between the fitted line of the interface layer and the fitted line of the bulk layer is calculated, and combined with the ratio of the maximum observation time to the mechanism switching time, a layering complexity intensity reflecting the degree of difference in medium layering is constructed. The hierarchical complexity intensity is mapped to a normalized fault criterion value using a logistic mapping function, the parameters of which are fixed through offline calibration. The fault criterion value is compared with a preset alarm threshold. When the fault criterion value exceeds the alarm threshold, a state diagnosis result is generated indicating that the insulation is damp or the interface is deteriorated.

Images